FAQ regarding our biostatistical consulting services and the analysis of clinical data
- Do I need a statistician/biostatistician?
- What is the difference between a statistician and a biostatistician?
- When is the best time to consult a statistician?
- How much prior notice do I need to give for a project?
- I have already collected my data; is it too late to consult a statistician?
- What if I want to perform my own analyses?
- I have already performed my statistical analysis; can you check it?
- What is involved in performing a statistical analysis?
- Will my data and/or research be kept confidential?
- Do you charge by the hour or per project?
- Do you offer a service subscription or retainer option?
- What happens with co-authorship?
- Do you assist with grant proposals?
Do I need a statistician/biostatistician?
Anyone who is conducting empirical research can benefit from the advice of a good statistician.
As statistics is a particularly specialised and technical field, academic supervisors will often recommend that honours and doctoral research students consult an expert when it comes to their statistical analysis. Similarly, clinicians and specialists in other areas will often save time and enhance the accuracy of their research when the advice of a biostatistician is sought and implemented. This is ever more important in a clinical context where uniform, high quality data is not always in abundance. A good statistician will be able to pick up on finer details pertaining to a statistical analysis, that may be overlooked by scientists and researchers specialising in other disciplines. Recent research shows that over the past 12 years data, including clinical data, has become increasingly complex and the methods available to best address today's research questions have increased in both precision and sophistication. It is therefore crucial to consult a biostatistician with the most up to date analytical techniques for your clinical niche.
What is the difference between a statistician and a biostatistician?
A biostatistician is a statistician who has received specialist training in the fields of medical and life sciences. This training usually includes skills important to medicine and public health such as epidemiological techniques and design of randomised controlled trials (Phase I, II and III clinical trials).
When is the best time to consult a statistician?
Consulting a statistician in the early, planning stages of your research can often save time and money later on. When a study is designed correctly this enables the best possible analysis to be performed that will extract the most out of your valuable data. It can also enable you to structure your study and data collection in such a way as to maximise statistical power.
In the set up stages of your study, a consultant statistician can calculate the optimal power to sample size ratio for your research budget. Knowing the experimental power that will likely be obtained with an intended sample size allows you to adjust sample size up and down depending on individual priorities.
- How much prior notice do I need to give for a project?
As biostatisticians may be working on multiple projects at a time it is best to contact us for around four weeks prior to your deadline, whether it be for a grand proposal, study design or final analysis. This will ensure we have ample scope to submit to you a draft and apply and adjust according to any feedback or evolving preferences. Shorter tasks such as sample size or power calculation can usually adhere to a much shorter time frame such as a few days - however in case of unforeseen circumstances or busy time periods, four weeks is a safe benchmark for prior notice.
I have already collected my data; is it too late to consult a statistician?
Statistical consulting can be beneficial at almost any stage of the research project. If data has already been collected, ready for analysis, the statistician will test statistical assumptions of the data in order to determine the most appropriate statistical methods to apply for this particular data and the hypotheses being tested (if any). The appropriateness of an analysis is influenced by various factors and often needs to be examined under a careful eye.
What if I want to perform my own analyses?
For individuals who wish to perform their own analyses and would like guidance to do this, I recommend an initial consultation to discuss your research aims and desired outcomes. If the data has already been collected, I will check the assumptions of this data in order to inform the most appropriate statistical approach. From here I can draw up a plan for analysis of which you can implement yourself. This will let you know what your options are for the analysis of this data so that you can get it right first time.
Alternatively, If you have an existing plan drawn up and just want me to check over it or provide software advice, I can do this also.
Where required I can assist statistically savvy individuals to perform their own analyses by developing code and/or troubleshooting advice for various software packages.
I have already performed my statistical analysis; can you check it?
Yes, I can. In order to check an analysis that has already been performed I will need to first check the data set for errors and correct data format. From there I will usually check that statistical assumptions are met for the tests that have been performed. This will usually give a good idea as to whether the analysis used was appropriate for the data and for the research question/s.
If required I can also assist with interpretation of results.
There have been a few occasions where I have been contacted by doctoral candidates who have had incorrect or non-ideal statistical analyses performed on their data (sometimes by offshore freelancers) which require checking over and amending. For those who do not relish in statistics, this can lead to considerable frustration as extra time must now be expended to complete their dissertation or research report.
In order to avoid this I recommend checking the credentials of potential contractors and discussing the analysis plan in depth before it is performed. A consultant should be able to justify why they have taken a particular approach.
What is involved in performing a statistical analysis?
There are several crucial stages to the actual analysis. The first stage generally consists of data checking, cleaning and formatting data ready for analysis. During this stage decisions can be made as to the treatment of missing values and imputation of missing data can be performed where required. Once data has been prepared for analysis (in a typical hypothesis testing scenario) the data will be assessed to determine how well it meets the assumptions for the particular statistical tests selected. This generally includes assessing normality, skew and homoscedacticity. Univariate and multivariate outliers should also be assessed and can be treated via a selection of methods. For data that severely violates statistical assumptions, a differing approach, such as non-parametric testing, may be deemed a more appropriate method to adequately evaluate the predetermined research hypotheses. At this point the main analysis can be performed and results evaluated.
This is a typical hypothesis testing scenario. Individual cases may differ.
Will my data and/or research be kept confidential?
All data and study information is kept 100% confidential. No information whatsoever will be shared with third parties. If a second consultant is required to work on the project, this will only occur with advance permission from the client. Encryption services are also available upon request.
Do you charge by the hour or per project?
No two projects are the same. Some projects can be completed with less fuss than anticipated others have requirements which evolve as the project progresses. As such I find I can offer the best value by charging per hour.
If a client is working within a particular project budget, I am happy to discuss what services I can afford to offer for that budget.
Do you offer a service subscription or retainer option?
Yes, we offer the option of a biostatistical service subscription for light, medium or heavy biostatistical consulting needs. This involves paying a monthly subscription fee in order to retain the services of a biostatistician for a certain amount of hours each month. The fee structure graduates in line with hours retained per month. See here for more details on the standard subscription tiers we offer. If these sit outside of your needs but you still feel that a retainer is appropriate for you, we are happy to negotiate a bespoke plan.
What happens with co-authorship?
Biostatisticians form an integral part of any clinical research project, from the design and planning stage, to data management and statistical analysis, to final write-up. As such, each contributing statistician should be considered for co-authorship alongside other contributing scientists. Co-authorship is to be negotiated, independent of funding considerations, based upon the statistician or statisticians’ proportional contribution to the paper or papers.
Do you assist with grant proposals?
Yes, we assist with many aspects of grant proposals including sample size and power calculations, clinical study designs, cost estimations for the biostatistical consulting component of the research, and in the development and analysis of pilot studies.